Journal notes, Project Flowcharts, comparing to USGS plots
GELLER Labs "Backyard Science"
Thoughts on a proton precession magnetometer design - a Proton Magnetometer Project. Build an Earth's field magnetometer.
The goal of this project is a low cost high performance proton magnetometer (a digital magnetometer) kit for amateur scientists to be able to accurately measure and monitor changes in the Earth's total magnetic F field and to observe geomagnetic storms. There is a regular daily (diurnal) variation in the Earth's magnetic field. During events related to solar activity, there can be sudden changes in the field (such as a sudden impulse) as well as large excursions in the field which can be more than ten times the regular diurnal variation caused by magnetic storms.
(be sure to hit refresh to pick up our latest changes and entries)
Saturday, June 5, 2010
Very early morning: PDF. The FDM magentometer continued to run normally while a line of severe thunderstorms with very frequent lightning just moved through the area from about 2:15 am to 3:15 am.
Overnight: PDF, TXT, USGS PDF last couple of days PDF.
Why does the FDM amplitude (the amplitude of the fundamental frequency as determined by the FDM method) vary with sensor temperature? PDF (black curve is 50 point moving average, very rougly about 80 degrees F to 75 degrees F during this period). With the constant current power supply, I do not think it is related to coil resistance as a function of temperature unless somehow influenced by the Q of the coil. Presently running on Prestone De-icer in a 125 mL (~4 Oz) tightly capped Nalgene plastic bottle. Is it fluid properties with temperature? Fluid pressure in the bottle? Not a problem, just one more curiosity.
Sunday, June 6, 2010
Overnight: PDF, TXT, USGS PDF, few days PDF.
Documentation Note: I am beginning to draw diagrams and to prepare documents for the project article. I will post some of these as I prepare them. Here is a flow chart that shows the auto-retry process PDF. The control program (presently LabView) makes a single measurement (detailed flow chart to follow). The digitized precession signal is sent to the FDM module (presently a windows executable) and returns the fundamental frequency of the precession signal, its amplitude, figure of merit (FOM), and a narrow band signal to noise ratio (S/N). The Labview control program reads these parameters (frequency, amplitude, FOM, and S/N) and following the flow chart filters the results for acceptable values. If the measurement was unsuccessful, the auto-retry process waits about five seconds and then triggers another new measurement. If the measurement was successful, the data enters the array of field data to be plotted (with a one measurement cycle delay) and the FDM magnetometer goes into a measurement cycle delay of about 2 minutes.
Late evening: PDF, USGS PDF. Interesting pattern seen across the continental United States (lower 48 states), with some variation from station to station. The shape of the disturbances were strikingly similar across most stations, however, there was less registration of the events at the PR, Guam an HI stations. It is a little difficult to compare with Alaska records since the vertical scales to the North are different (100s of nT) from most of the lower 48 graphs (10s of nT). The relatively fast event between 6 pm and 7 pm (~2210 to 2230 UTC) was just captured in full shape with our 2 minute measurement rate, probably also including some number of missed points due to auto-retries, but it was recorded in both positive and negative extent. The slower impulse between 9 pm and 10 pm (~0115 to 0150 UTC) was captured perfectly. The two minute sample rate could be reduced to say, 1 minute, however for now, the slightly slow measurement rate gives an easier to read plot when viewed from one to several days. I wonder if there should there be an "automatic transmission" were the magnetometer would sample faster when it detects large changes (relatively high dB/dt for say more than one to three points)? Or, maybe variable rate sampling would make the graph more difficult to interpret visually? I suppose a higher rate could run for a fixed period, say 30 minutes from a series of high dB/dt data points.
Monday, June 7, 2010
Overnight: PDF, TXT, USGS PDF.
The trend of higher FDM amplitudes for lower sensor temperature continues PDF. (The outdoor temperature at the sensor fell off to about 43 degrees F last night.) This FDM amplitude dependence on sensor temperature is just a curiosity and has no affect on the determination of the precession frequency or the corresponding geomagnetic field computed by the FDM method.
Tuesday, June 8, 2010
Overnight: PDF, TXT, USGS PDF. Ran overnight with a one minute measurement delay. That timing gives too dense an overnight plot, so returning to a two minute delay. As discussed earlier, it might be worth investigating a routine to decrease the measurement time for interesting events.
Drawings: Here is a flow chart of the overall measurement cycle: PDF. The FDM executable performs the filter diagonalization method (See: Jan 21, Jan 23 entries) , an ultra high resolution (1 milli Hz) frequency estimator borrowed from the NMR spectroscopy and physical chemistry community, (Vladimir A. Mandelshtam, Howard S. Taylor, Harmonic inversion of time signals and its applications, Journal of Chemical Physics (1997), Volume 107, Issue 17, 1997, Pages 6756-6769). FOM is the figure of merit generated by the FDM algorithm. The narrow band signal to noise ratio (S/N) is computed by the square root of the peak amplitude of the FID (free induction decay or precession signal) waveform over the sum of the squares of n spurious noise frequencies of all noise sources including those related to AC line noise. A separate block diagram to follow will show the hybrid FET-Relay timing sequence and the precession waveform filtered envelope computation (see June 17 entry) in more detail.
Sunday, June 13, 2010
Overnight: PDF, TXT, USGS PDF. Offsets are caused by the usual vehicles. Generally a small impulse can be distinguished from an arrived vehicle (depressed or lower field) or a departed vehicle (raised or higher field) by the number of points in the transition. Vehicles tend to cause step like transitions, where as small impulses tend to have two or more points within the slope of the changing field.
Tuesday, June 15, 2010
Last few days: PDF, TXT, USGS 24 hour PDF.
Wednesday, June 16, 2010
Last couple of days: PDF, 24 hour PDF, TXT, USGS 24 hour PDF, a good contrast of a relatively quite geomagnetic day with a somewhat unsettled day (albeit with a limited range of field swing and at sub storm levels).
Afternote: In later months, we learned that our magnetograms follow very closely (substantially identical at times) the Natural Resources Canada (NRCan) Ottawa (OTT) total field magnetograms. OTT is about 200 miles to our north. It is common place for our magnetograms to be quite different from the USGS FRD station, 400 miles to our south west. When visually comparing magnetograms, keep in mind that small changes in vertical scale resolution can give "visual" affects that might make very similar or even idential magnetograms (graphs of the total field value or "F scalar") look quite different.
On comparing our geomagnetic field plots to the USGS plots: Since we have the luxury of reporting only a single station, our PDFs use significantly higher vertical (y axis) resolution than the USGS plots. Every once in a while, even though we have done many successful comparisons, I compare the two and wonder about the correlation. Take this evening's plots for example, our typical expanded plot PDF compared to the USGS plot PDF. But, then reducing our vertical scale gives a pretty good visual match PDF, (with the exception of our momentarily stopped 3 pm school bus, and the small offset between very roughly 2:20 pm and 2:40 pm caused by a parked vehicle slightly depressing the field during that time).
Also, when comparing to USGS plots, consider that (for the nearest USGS observatories): we are roughly 400 miles (~650 km) NNE (north north east) of the Fredericksbug (Corbin or FRD), VA station, and roughly 1300 miles (~2100 km) NE of the Stennis, MS BSL observatory. Yet, as we have noted in days past, some events, such as for example sudden impulse events, often give very similar wave shapes across a large number of observatories making for good comparison events.
Thursday, June 17, 2010
Overnight: PDF, small vertical axis scale PDF, TXT, USGS PDF. Continuing the comparison to USGS data this morning, the Corbin, VA site had a vertical scale on our (landscape) screen of about .6" to 44 nT or roughly 76 nT / 1". Our (portrait) display was set to about 13 nT / 1". Our vertical scale is generally set to 5 nT per vertical division, however the number of divisions varies from day to day. We compress the vertical scale (or use a lot of vertical divisions) during relatively large field excursions, such as during a magentic storm. We made another comparable graph today for visual comparison.
Interesting to note that we did not see much of the typical 8-10 am to afternoon diurnal variation cycle today (at least not the part before noon) (there where two relatively small ~ +2 nT offsets from arrived parked vehicles), while the southern Corbin, VA trace does show the typical cycle. I think such differences hint at the somewhat limited use of relatively distant observatories in confirming proper magnetometer operation. On the other hand, as mentioned yesterday, general comparisons over many days as well as impulse events that appear at a number of observatories can be quite helpful for verification purposes. We do seem be seeing now at least a part of the afternoon diurnal cycle: PDF.
More Drawings: Here is a flow chart PDF of the algorithm that generates the filtered envelope display of the precession signal. This function very roughly produces a FID (free induction decay or precession signal) filtered envelope similar to those we produced earlier in hardware using the hp 3581A/C waveform analyzer using the Tektronix 2440 scope as the waveform display JPG1 JPG2. We begin by taking the absolute value of all points in the FID array. Next, there is a moving RMS (root-mean-square) computation (similar to a box car operation) of M points (we are presently using 600) of the N points of the prescession waveform (presently 11,000). The box advances one point, and the computation is repeated. The box continues to advance one point at a time until the last of the N points is reached (it probably should be N-600, so there is always a full M calculation).
Project Documentation (very early stages)
QUESTIONS/COMMENTS/notice of typos, etc. send email to joegeller @ gellerlabs dot com
COPYRIGHT © 2009, 2010 JOSEPH M. GELLER, All rights reserved.